Grading and Reporting Student Learning Follow Up

Breakout session with Thomas R. Guskey
A question and answer session following his keynote on grading and reporting student learning


This breakout session consisted of Guskey responding to audience questions about his keynote. Some relevant conversations: 

The ACT cannot be used as a measure to evaluate your educational program.
As outlined in his keynote, Guskey states that the grades that high schools report have no meaning. A 4.0 GPA from one high school means something totally different than a 4.0 GPA from another. Colleges cannot rely on a GPA to determine a student's knowledge. Enter the ACT - this test is NOT designed to assess student skills; it is designed solely to create a bell curve, a spread, a normal distribution to help colleges make distinctions between students for college entrance. Even if there is a quality question on the ACT that accurately measures a worthwhile skill, if everyone teaches this skill and students start getting this question right, the next year this question would be removed from the test. The ACT is scored on a curve, which goes against the purpose of grading. A test graded on a curve fails 50% of the students, regardless of what they have learned. This is not the appropriate test to measure student learning.

High percentages are not the same as high standards.
Let's say I decide that students need to score an 80% or above to demonstrate proficiency. This style of percent cutoff is completely unreliable because of the variability in the design of assessments. An example to illustrate how assessments are unreliable: here's a low-level but difficult question:

Who was the 17th president of the United States?

Fewer than 10% of students can answer this correctly. However, consider the following alteration: 

Who was the 17th president of the United States?

• Abraham Lincoln
• Andrew Jackson
• Ulysses S. Grant
• Millard Fillmore 

Now 30% of students can answer this correctly (obviously you have a 25% chance of guessing the right answer). But what about this alteration?

Who was the 17th president of the United States?

• George Washington
• Andrew Jackson
• Jimmy Carter
• Bill Clinton 

Now 60% of students can answer this correctly. Notice that the format - multiple choice - stayed the same, but the choices can give away the answer. One last alteration: 

Who was the 17th president of the United States?

• The War of 1812
• Andrew Jackson
• The Louisiana Purchase
• A Crazy Day for Sally 

Now 90% of students can answer this correctly. So what's the point? The point is that although each question assessed the exact same learning objective, there was a huge variation in difficulty. It would be unreliable and unfair to use the same percentage cutoffs for these 4 questions. Focusing solely on percentage cutoffs is a huge mistake - the easiest way to increase those percentages is to make the assessment easier (which as shown in this example can be done without adjusting the targeted skills!).  

The first grade you give sets the tone for the student for the rest of the year.
When a student starts a new class, they have a fresh start - it's a new teacher, it's a new year, maybe this one will be different! But when they get back their first quiz grade of the quarter and it's a C, this sets the tone for the rest of the year. The student decides that they are a C student, and on the next quiz, they expect to get a C, and they work accordingly. The first grade you give to a student is the most powerful. Do not make your first assessment impossibly difficult just to set the tone for work ethic and discipline. 

A novel homework policy - no individual homework grade, but adds 5% to next quiz.
The basic idea is to flip student's concept of homework - they are far more willing to do it if it is presented as a reward (5 bonus percent on the next quiz) as opposed to a punishment (drill and kill grunt work). In particular, they are far more likely to do it in a timely fashion! Although the idea of giving 5% extra credit to a quiz may bother some, it is more or less mathematically equivalent as having the traditional homework grade and averaging them together. 

You should always give students a second chance with assessments. 
Some people are against this - the typical argument is this: "A surgeon doesn't get a second chance on an operation! A pilot doesn't get a second chance to land a plane!" This is true; however, that surgeon isn't operating on a person for the very first time, nor is that pilot landing his or her very first plane. They have had thousands of hours of practice before that high stakes experience. When you teach your content skills, it is the first exposure your students are getting with the skill. So they should be afforded more than one chance to practice and prove they have those skills.

Peter's opinion:
I was really surprised to hear some of the things this guy was saying! He actually teaches statistics, so he has  a particular clarity on the meaninglessness of percents.  Some eye-opening stuff. He didn't really get a chance to get in to some more specific practical implementations. For instance, I agree with the notion that you should always give second chances to prove knowledge, but if a student knows this, why bother studying for the first quiz? Or the second one for that matter? Still, I feel like this workshop provided some good ideas to begin implementing with my own grading scheme...